Separation and Quantification of Four Main Chiral Glucosinolates in Radix Isatidis and Its Granules Using High-Performance Liquid Chromatography/Diode Array Detector Coupled with Circular Dichroism Detection

As chemical drugs, separation and quantification of the specific enantiomer from the chiral compounds in herbal medicines are becoming more important. To clarify the chemical characterization of chiral glucosinolates—the antiviral active ingredients of Radix Isatidis, an optimized efficient method of HPLC-UV-CD was developed to simultaneously separate and quantify the four main chiral glucosinolates: progoitrin, epiprogoitrin, and R,S-goitrin. The first step was to determine progoitrin, epiprogoitrin, and R,S-goitrin using HPLC-UV, and then determine the R-goitrin and S-goitrin by coupling with CD detection. Subsequently, through the linear relations between anisotropy factor (g factor) and the percent optical purity of R-goitrin, the contents of R-goitrin and S-goitrin from the R,S-goitrin mixture were calculated separately. Furthermore, the chemical composition features of the four chiral glucosinolates in 37 samples from crude drugs, decoction pieces, and granules of R. Isatidis were conducted. The total content of the four glucosinolates was obviously higher in crude drugs, and the variance character of each glucosinolate contents was different. In summary, the accurate measurement method reported here allows for better control of the internal quality of R. Isatidis and its granules and provides a powerful approach for the analysis of other chiral components in traditional Chinese medicines.


Introduction
Natural products from traditional Chinese medicines (TCMs) represent a major part of today's pharmaceutical market as they possess a number of biological activities, such as antimicrobial, antitumor, anti-inflammatory, antiviral, cardiovascular activities, and so on [1]. Since chirality is a fundamental characteristic of nature, a large number of well-known therapeutic ingredients from elution were compared. Comparison with different compositions of acetonitrile/methanol-water, methanol-water with different modifiers including formic acid and ammonium acetate showed the combination of methanol and 30 mM ammonium acetate (adjusted pH for 5.0 with formic acid) gave an optimal mobile phase system for glucosinolates separation. As shown in Figure S1, in order to acquire better peak intensity, five UV wavelengths-230, 242, 245, 254, 280 nm-were detected and 230 nm gave the best results. In addition, the maximum absorption CD wavelength of Rand S-goitrin was focused on 195 and 280 nm, respectively, whereas the detection range of the present CD-2095 detector was 220-490 nm. Therefore, the CD wavelength was set at 280 nm.

Method Validation of Quantitative Analysis
As shown in Table 1, all calibration curves of the glucosinolates showed good linearity (R 2 > 0.9991) within a relatively wide range of concentration (0.0365-3.080 mg/mL). The limit of detection (LOD) and the limit of quantification (LOQ) were lower than 0.802 and 2.305 µg/mL, respectively. Table 1. Regression equations, linear range, limit of detection (LOD) and limit of quantification (LOQ) for three analytes. The known purities of R-goitrin (%) and S-goitrin (%) were mixed and simultaneously determined by UV and CD to calculate the g factors ( Table 2). The relative purities of R-goitrin (%) and g factors made calibration curves. The calibration curve between g factors (X 2 ) and the different relative purities of R-goitrin (Y 2 ) was Y 2 = −0.0092X 2 + 0.4965 (R 2 = 0.9970), which showed a good linearity with a wide range of R-goitrin purities (0~100 %).

Glucosinolates
The precision, repeatability, and stability of the developed method were also validated for each analyte, and the RSD values were less than 3.0%. The recoveries of five glucosinolates were measured from 99.1 to 103.3%, and the RSD values were less than 3.0% (Table 3).

Glucosinolates Profiles of Radix Isatidis and Its Granules
Compared to previous published methods [19][20][21][22], glucosinolates profiles of crude drug, decoction pieces, and granules of R. Isatidis were performed using HPLC-UV-CD, which improved the limitation of high-cost chiral columns, complex operation of sample preparation, and simultaneous quantification of the characteristic glucosinolates. In the typical UV and CD chromatograms of R. Isatidis samples (Figure 1), R-goitrin and S-goitrin showed the same retention time, UV absorption feature, and overlapping peak. They could not be separated by UV detection without using the specific chiral columns ( Figure 1A,B; upper). In contrast to that, R-goitrin and S-goitrin had completely opposite chromatographic profiles at the same retention time as shown in CD chromatograms ( Figure 1A,B; lower). Notably, in the typical CD chromatograms, only R-goitrin could be detected in the mixed R,S-goitrin, crude drugs, decoction pieces, and granules of R. Isatidis. Therefore, the R-goitrin content could be quantified using CD detector, while the S-goitrin content was calculated using the mixed R,S-goitrin and the determined R-goitrin content. Combining the CD chromatographic properties with total content of the mixed R,S-goitrin, it provided an efficient method to determine the content of R-goitrin in the mixed R,S-goitrin without chiral separation.

Glucosinolates Profiles of Radix Isatidis and Its Granules
Compared to previous published methods [19][20][21][22], glucosinolates profiles of crude drug, decoction pieces, and granules of R. Isatidis were performed using HPLC-UV-CD, which improved the limitation of high-cost chiral columns, complex operation of sample preparation, and simultaneous quantification of the characteristic glucosinolates. In the typical UV and CD chromatograms of R. Isatidis samples (Figure 1), R-goitrin and S-goitrin showed the same retention time, UV absorption feature, and overlapping peak. They could not be separated by UV detection without using the specific chiral columns ( Figure 1A,B; upper). In contrast to that, R-goitrin and S-goitrin had completely opposite chromatographic profiles at the same retention time as shown in CD chromatograms ( Figure 1A,B; lower). Notably, in the typical CD chromatograms, only R-goitrin could be detected in the mixed R,S-goitrin, crude drugs, decoction pieces, and granules of R. Isatidis. Therefore, the R-goitrin content could be quantified using CD detector, while the S-goitrin content was calculated using the mixed R,S-goitrin and the determined R-goitrin content. Combining the CD chromatographic properties with total content of the mixed R,S-goitrin, it provided an efficient method to determine the content of R-goitrin in the mixed R,S-goitrin without chiral separation. As shown in Figure 1, the chemical composition of progoitrin, epiprogoitrin, and R,S-goitrin in crude drugs, decoction pieces, and granules of R. Isatidis showed that the chromatographic patterns were found to be consistent between all 37 samples but their contents obviously varied. The three analytes could be easily detected in the UV chromatograms, while only R-goitrin was determined in CD chromatograms. We think the different absorption intensity between UV and CD detection might be due to the differences in chemical structures. Therefore, the content variations of these characteristic glucosinolates should be conducted.

Chemical Comparison on the Basis of Contents of Four Glucosinolates
The contents of four glucosinolates-progoitrin, epiprogoitrin, R-goitrin, and S-goitrin-were simultaneously quantified in 37 samples using HPLC-UV coupled with CD detection. As shown in Figure 2 and Table S2, the total contents of four glucosinolates in three types of R. Isatidis were significantly different. It was obviously higher in crude drugs (0.77-17.54 mg/g, Mean ± SD: 5.04 ± 5.03) and decoction pieces (0.56-8.29 mg/g, Mean ± SD: 2.72 ± 3.07), while the glucosinolates occupied relatively lower contents in its granules (0.03-0.87 mg/g, Mean ± SD: 0.21 ± 0.24). As shown in Figure 1, the chemical composition of progoitrin, epiprogoitrin, and R,S-goitrin in crude drugs, decoction pieces, and granules of R. Isatidis showed that the chromatographic patterns were found to be consistent between all 37 samples but their contents obviously varied. The three analytes could be easily detected in the UV chromatograms, while only R-goitrin was determined in CD chromatograms. We think the different absorption intensity between UV and CD detection might be due to the differences in chemical structures. Therefore, the content variations of these characteristic glucosinolates should be conducted.

Chemical Comparison on the Basis of Contents of Four Glucosinolates
The contents of four glucosinolates-progoitrin, epiprogoitrin, R-goitrin, and S-goitrin-were simultaneously quantified in 37 samples using HPLC-UV coupled with CD detection. As shown in Figure 2 and Table S2, the total contents of four glucosinolates in three types of R. Isatidis were significantly different. It was obviously higher in crude drugs (0.77-17.54 mg/g, Mean ± SD: 5.04 ± 5.03) and decoction pieces (0.56-8.29 mg/g, Mean ± SD: 2.72 ± 3.07), while the glucosinolates occupied relatively lower contents in its granules (0.03-0.87 mg/g, Mean ± SD: 0.21 ± 0.24). Most previous studies have focused only on the content of R,S-goitrin mixture in R. Isatidis and its related products. In the present study, R-goitrin accounted for a majority of the R,S-goitrin component in crude drugs, decoction pieces, and granules of R. Isatidis; the R-goitrin content was twice as much as that of S-goitrin. The results might provide positive evidence that R-goitrin is responsible for the pharmacological properties of R. Isatidis.
It was noticed that progoitrin and epiprogoitrin contents showed an obvious declining trend in crude drugs and decoction pieces of R. Isatidis. In particular, they could not be detected in most of the granule samples. Compared with crude drugs, R-goitrin and S-goitrin contents in decoction Most previous studies have focused only on the content of R,S-goitrin mixture in R. Isatidis and its related products. In the present study, R-goitrin accounted for a majority of the R,S-goitrin component in crude drugs, decoction pieces, and granules of R. Isatidis; the R-goitrin content was twice as much as that of S-goitrin. The results might provide positive evidence that R-goitrin is responsible for the pharmacological properties of R. Isatidis.
It was noticed that progoitrin and epiprogoitrin contents showed an obvious declining trend in crude drugs and decoction pieces of R. Isatidis. In particular, they could not be detected in most of the granule samples. Compared with crude drugs, R-goitrin and S-goitrin contents in decoction pieces increased slightly. The traditional processing and extraction methods could improve the biotransformation of progoitrin and epiprogoitrin, and then increase the content of the degradation products (R-and S-goitrin). This is consistent with our previous report [16,17]. However, the content dynamic change features of the glucosinolates in crude drugs, decoction pieces, and granules as well as the main influencing parameters for biotransformation among these glucosinolates need to be further investigated.

Sample Materials
Fifteen crude drugs (B1-B15), nine decoction pieces (B16-B24), and thirteen granules (B25-B37) of R. Isatidis were collected from the main commercial herbal markets and different herbal manufactories in China (Table S1). The vouchers have been deposited in Shanghai R&D Centre for standardization of Chinese Medicines, Shanghai University of Traditional Chinese Medicine.

Chemicals and Reagents
The four glucosinolates-epiprogoitrin, progoitrin, and R,S-goitrin-were isolated and purified from the root of I. indigotica Fort. in our laboratory. R-goitrin and S-goitrin were prepared from R,S-goitrin using the Shiseido ® CD-pH chiral column (5 µm, 250 × 4.6 mm) with acetonitrile-water (30:70, v/v). Their chemical structures ( Figure 3) were elucidated by a series of spectroscopic and chemical analyses [23]. The purities of five glucosinolates were determined to be higher than 98.0% through HPLC-DAD analysis. HPLC-grade acetonitrile, methanol, formic acid (≥98.0%, HPLC grade), and ammonium acetate (≥98.0%, HPLC grade) for HPLC analysis were purchased from Thermo Fisher Scientific (Swedesboro, NJ, USA). Water was prepared by a Milli-Q ® system (Millipore, MA, USA). All other reagents of analytical grade for extraction were purchased from Tianjin Damao Chemical Reagent Factory (Tianjin, China).
Molecules 2018, 23, x FOR PEER REVIEW 6 of 9 pieces increased slightly. The traditional processing and extraction methods could improve the biotransformation of progoitrin and epiprogoitrin, and then increase the content of the degradation products (R-and S-goitrin). This is consistent with our previous report [16,17]. However, the content dynamic change features of the glucosinolates in crude drugs, decoction pieces, and granules as well as the main influencing parameters for biotransformation among these glucosinolates need to be further investigated.

Sample Materials
Fifteen crude drugs (B1-B15), nine decoction pieces (B16-B24), and thirteen granules (B25-B37) of R. Isatidis were collected from the main commercial herbal markets and different herbal manufactories in China (Table S1). The vouchers have been deposited in Shanghai R&D Centre for standardization of Chinese Medicines, Shanghai University of Traditional Chinese Medicine.

Chemicals and Reagents
The four glucosinolates-epiprogoitrin, progoitrin, and R,S-goitrin-were isolated and purified from the root of I. indigotica Fort. in our laboratory. R-goitrin and S-goitrin were prepared from R,S-goitrin using the Shiseido ® CD-pH chiral column (5 μm, 250 × 4.6 mm) with acetonitrile-water (30:70, v/v). Their chemical structures ( Figure 3) were elucidated by a series of spectroscopic and chemical analyses [23]. The purities of five glucosinolates were determined to be higher than 98.0% through HPLC-DAD analysis. HPLC-grade acetonitrile, methanol, formic acid (≥98.0%, HPLC grade), and ammonium acetate (≥98.0%, HPLC grade) for HPLC analysis were purchased from Thermo Fisher Scientific (Swedesboro, NJ, USA). Water was prepared by a Milli-Q ® system (Millipore, MA, USA). All other reagents of analytical grade for extraction were purchased from Tianjin Damao Chemical Reagent Factory (Tianjin, China).

Preparation of Standard Solutions
Each of the five reference compounds was accurately weighed and then dissolved in water to prepare the standard solutions of 2.084 mg/mL for progoitrin, 3.080 mg/mL for epiprogoitrin, 1.168 mg/mL for R,S-goitrin, 0.5488 mg/mL for R-goitrin, and 0.5080 mg/mL for S-goitrin. For UV detection, a series of standard solutions were prepared by appropriate dilution of the stock solutions to make calibration curves. The calibration curve was obtained by plotting peak area (Y 1 ) of each reference glucosinolate against the concentration of the corresponding compound (X 1 ). Furthermore, the known purities of R-goitrin (%) were determined by UV and CD to calculate the anisotropy factor (g = ∆A/A). The calibration curve depended on g factors (Y 2 ) of each reference R-goitrin against the relative purities of R-goitrin (X 2 ).

Preparation of Sample Solutions
Crude drugs and decoction pieces: The respective sample was pulverized to obtain homogeneous fine powder. 1.0 g of the fine powder was accurately weighed and extracted with 30 mL water by refluxing in a water bath for 1 h, then cooled and filtered. 1 mL of the filtrate was dissolved with 1 mL water containing 2% formic acid (v/v). The mixed solution was packed into a SPE column (Waters Oasis WAX 3cc Cartridge/ 60 mg, Waters, USA), which was activated with 3 mL methanol and then washed with 3 mL water before adding sample solution. The SPE column was orderly eluted with 3 mL water containing 2% formic acid, 2 mL methanol (A) and 2 mL methanol containing 5% ammonium hydroxide (v/v, B). Subsequently, the eluted solutions of A and B was dried by flushing with nitrogen (N 2 ). The residue was dissolved with 0.1 mL methanol for HPLC analysis.
Granules: The granules were pulverized and screened through a 300 µm sieve to obtain homogeneous fine powder. 2.0 g (containing sugar) and 1.0 g (no sugar) of the fine powder was accurately weighed and extracted with 10 mL water by ultrasonication at room temperature for 15 min, then cooled and filtered. 1 mL of the filtrate was dissolved with 1mL water containing 2% formic acid (v/v). The mixed solution was packed into a SPE column. Subsequently, the sample solution was prepared through the same operation procedure with sample solutions of crude drugs and decoction pieces.

Method Validation
Linearity was assessed by generating six-point calibration curves for each reference compound. The precision was evaluated by replicate injections of a mixture solution containing four glucosinolates and a sample solution of R. Isatidis (B10). Six injections of the same preparation solutions per day were investigated for three days. Six preparation solutions of a sample (B10) were analyzed for repeatability. To determine the recovery rate of extraction, the four reference components (approximately 50%, 100% and 150% of the original amount) were added to the weighed powder of R. Isatidis that was extracted and analyzed.

Statistical Analysis
As the CD signal depends only on the enantiomeric composition of the chiral molecule whereas UV absorbance is related to the analyte concentration, the measurement method of the enantiomeric content from the chiral sample was referred to the reported literatures [24,25]. Firstly, the CD detector recorded both dichroic (∆ε or ∆A) and UV (ε or A) signals at the optimized wavelength and calculated the anisotropy factor (g = ∆ε/ε or ∆A/A), which is dependent of the enantiomeric purity. Secondly,